People spend about 7% of their time eating. That’s almost twice as much as the percentage of time cars are used. It’s an amazingly small number, especially when you consider that a vehicle is the second most expensive asset, after a home, that most people purchase.

Source: McKinsey

Even when we are using our cars, the vast majority of trips have only one passenger (on average 1.6 persons/vehicle mile) and this is especially true for work trips where the average is even lower (1.1 persons/vehicle mile). That means that even for the 4% of the time that a vehicle is actually in use, only about a third of the vehicle’s capacity is being used — and that’s not including unused trunk space.

Since there are about a billion vehicles that exist in the world today, the underutilization is vast. You may object that people don’t use their homes or coffee makers much of the time either. The tricky thing about cars is that they also require vast amounts of infrastructure to support their non-use.

During the 96% of the time a car sits idle, dripping fluids, rusting and slowly depreciating, it needs a place to be.
‍
Unfortunately, a car doesn’t just require one larger-than-car sized parcel of real estate to be set aside for it. We need a place to park at home and at work, outside Safeway and near to the gym or the hair stylist. For each of America’s more than 250 million cars, about four parking spots are set aside. You can do the math.

Note: This estimate assumes that the amount of space set aside for car parking is just the size of the parking space itself. In reality, parking lots require significant amounts of additional space for cars to enter, exit and maneuver, meaning that the true footprint is likely significantly larger than represented.

But the tremendous cost of all this parking is hidden from those using (or mostly not using) all of these parking spots since drivers almost always park for free. Indeed, real estate includes the cost of parking bundled with it in the form of parking minimums for new developments. Take a look for example at this summary of Houston’s parking requirements: 1.333 spots are required for every one bedroom apartment and 5 spots per green at a golf course. As Donald Shoup describes, coming up with these minimums is as rigorous a science as bloodletting.

Once a vehicle ventures out from a parking bay onto the road, another set of inefficiencies emerge. Roads cost money: the Federal Government spends about $100B annually on infrastructure, half of which goes to highways. Meanwhile, state and local governments spend vastly more.

Source: Urban Institute

The taxes to pay for roads are taken from everyone, but only those who can afford car ownership get to benefit from these investments. Since road costs aren’t paid directly by users (except for a small minority of toll roads, which are generally unpopular) people think of road utilization — like parking — as effectively “free”. In reality, government at all levels is massively subsidizing car ownership.

People aren’t so great at using these roads. They cut each other off and drive at a variety of different speeds. They need large following distances, fixed lanes and traffic lights to help coordinate their actions at intersections.

Source: McKinsey
‍US city land allocations

If you compare the movement of vehicles along a road with the movement of Kiva robots along the floor of an Amazon warehouse, there’s a big difference: cars slosh around like fluids in a pipe whereas kiva robots move like data packets in network. In other words, the one is governed by analog dynamics, the other by a digital system. This difference matters when roads are congested. System level interactions mean that roads are least effective when we need them most.

Congestion is expensive. According to a 2019 Texas A&M’s Transportation Institute report, urban congestion causes Americans to travel an additional 8.8 billion hours while burning an extra 3.3 billion gallons (12.5 billion liters) of fuel, costing the economy an estimated $179 billion.

And then there are traffic accidents. People drive while sleepy, intoxicated, texting - or all at the same time. Sometimes even when people are fully focused they make slow or bad judgements. Every year around 40,000 Americans die in vehicle accidents - the leading cause of death amongst young adults. Large as this number is, it is dwarfed by the 1.25 million people who die globally each year in car crashes, 90% of whom live in low- or middle-income countries. And this carnage comes with significant costs: according to NHTSA, when coupled with injuries (around 4 million each year), damage to property and accident-related congestion, direct costs of crashes in the US amounted to $242 billion in 2010 and ballooned to $836 billion when quality of life impacts were factored in.

Given these costs, it is surprising how little information we have about what vehicles are doing on the road at any given time. For the most part, we only have a one-dimensional picture of how well or badly people are driving, how traffic flows are changing, what behaviors are creating slowdowns, etc. Much data of the data we have is collected by conducting surveys. And the data about fatalities and accidents (cited above) takes months or years to be released and has very sparse information about non-fatal accidents. Road planners (and to some extent insurance companies) are like cardiologists making diagnoses without the benefit of an EKG.

For the rare moments when cars are in use, you might hope that they at least run efficiently. Alas, the small power plant housed in most vehicles, which creates noise, heat, vibration and a range of pollutants as it blows up fuel to generate energy, loses 86% of that energy before it reaches the wheels.

Source: McKinsey

Even the energy moving the wheels is often wasted. Besides drivetrain inefficiencies, the size and shape of vehicles adds to the fuel they burn. All those empty seats mentioned earlier need to be moved around. Cars also have lots of extra metal to protect against crashes, adding weight. And people drive inconsistently, burning extra energy as they go. American vehicles have a combined capacity to produce ten times as much energy as the country’s power stations, but most of that capacity is not being used most of the time and wasted when it is.

On top of that, 92% of America’s transportation energy needs are powered by a single fuel source, petroleum, the price of which is still largely dictated by the OPEC cartel. Our climate is affected by the massive volume of gases spewed from vehicles' engines into the air. And there are direct impacts on our health: vehicles are a major source of outdoor air pollution to which the OECD attributes more than 3 million premature deaths per year; more Americans die from car emission related pollution each year than in traffic crashes.

Carmakers are focused on powertrains and engines even as their importance declines in satisfying the needs and desires of consumers. As Tony Fadell, creator of the iPod and Nest, noted: “Detroit’s approach to engines is as if every computing device company had to recreate the Intel chip to be able to build their products — a lot of very heavy lifting for very little differentiation.”

When it comes to the digital, carmakers have added silicon to vehicles in the way they might add any other component in order to enable additional functionality: Electric windows? Add a chip to run the motor. More fuel efficiency needed? Add another chip to manage the engine. And so on. All of these computers are connected through an internal network called the CAN bus that was developed in the 1980s.

These computers are housed in large, clunky, silver boxes called electronic control units (ECUs) which are scattered around the vehicle. If you’ve ever played Clue, Colonel Mustard with an ECU in the Ballroom could be a plausible solution.

However, digital systems don’t scale well in the way carmakers are using them. The current network architecture is not only underutilizing computing resources by adding additional processors rather than sharing the same resources for similar tasks, but the cabling connecting them and the metal surrounding these silver boxes is adding significant weight and complexity to the manufacturing process. Hardware is being added to do what should be done using software and a reimagined network architecture.

A direct consequence of the haphazard way in which cars have been networked is their cybersecurity vulnerability. At some point, it became apparent that connecting vehicles to the internet would offer various benefits: telematics, infotainment and software updates. Cars collect a lot of data and they need a channel to move it through. However, on the system level, cars have not been networked with security in mind and so they are rather susceptible to hacks.

Consider: the iPhone is updated (often completely reinvented in the process) every twelve months. In the decade since its creation, this meant about 10 major hardware updates. In contrast, cars are redesigned on average every 5 years. Consequently, your phone has been updated about as many times in the last decade as your car over the last half century.

This is just the hardware update cycle. The iPhone’s software is updated every few months while the apps running on it are often updated every week. Aside from Tesla, carmakers don’t really do software updates (other than patches that you get during servicing). If you want to update the software yourself, good luck figuring out how to plug a USB drive into your Ford. This is a problem given the relatively long lifespan of vehicles. This problem is partly tied to the network architecture of vehicles which have been slow to adapt to digitization. And partly it is linked to how carmakers distribute vehicles, deferring servicing to their dealership networks which makes them reticent to bypass them through OTA software updates.

Fast update cycles work like interest: they add extra value by compounding over time. Conversely, the second-order inefficiency of slow update cycles compounds the other inefficiencies across the entire transportation system.

Time is money as the saying goes and a person is spending their time when they are driving a car (although once again, most people think of this activity as "free", in contrast to say a taxi where you have to pay someone else). In economic terms, this is called opportunity cost and it adds up to quite a lot: The average American’s income is just over $26/hour and she spends about 48 minutes driving each day. That means that the value of daily time given over to driving is just over $20 or about $7,600 a year per person. That’s more than $1 trillion worth of labor invested into driving when multiplied across the American workforce.

If people were not driving, they wouldn’t necessarily be productive. But a trillion dollars multiplied by even a small fraction is still a big number. And if people aren’t working, then they’d be doing something else they’d rather be doing (relaxing, sleeping, watching Mean Girls, playing Fruit Ninja, etc). This time consumers invest into driving their cars has been taken for granted by carmakers and society, in large part because until now, you haven’t had much of a choice or had much reason to think about it. But things are changing.

Share this chapter with others: